Anda di halaman 1dari 16

On the absence of asymmetric wakes for periodically plunging finite wings

D. E. Calderon, D. J. Cleaver, I. Gursul, and Z. Wang



Citation: Physics of Fluids (1994-present) 26, 071907 (2014); doi: 10.1063/1.4891256
View online: http://dx.doi.org/10.1063/1.4891256
View Table of Contents: http://scitation.aip.org/content/aip/journal/pof2/26/7?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in
Control of a jet-in-cross-flow by periodically oscillating tabsa)
Phys. Fluids 24, 055107 (2012); 10.1063/1.4719150

Investigation of the effect of external periodic flow pulsation on a cylinder wake using linear stability analysis
Phys. Fluids 23, 094105 (2011); 10.1063/1.3625413

Finite time Lagrangian analysis of an unsteady separation induced by a near wall wake
Phys. Fluids 22, 075103 (2010); 10.1063/1.3459154

Jet switching phenomenon for a periodically plunging airfoil
Phys. Fluids 19, 027104 (2007); 10.1063/1.2565347

Shear layer oscillation along a perforated surface: A self-excited large-scale instability
Phys. Fluids 14, 4444 (2002); 10.1063/1.1519531


This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
PHYSICS OF FLUIDS 26, 071907 (2014)
On the absence of asymmetric wakes for periodically
plunging nite wings
D. E. Calderon, D. J. Cleaver, I. Gursul, and Z. Wang
Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, United Kingdom
(Received 14 March 2014; accepted 14 July 2014; published online 31 July 2014)
It has previously been shown that, at high Strouhal numbers, oscillating airfoils
can produce deected jets that can create very high lift-coefcients for otherwise
symmetric scenarios. These deected jets form through pairing of the trailing-edge
vortices to create asymmetric vortex couples that self-propel at an angle to the
freestream, resulting in an asymmetric ow eld and non-zero lift. In this paper
results are presented that indicate these high-lift deected jets cannot form for nite
wings. Instead of the straight vortex tubes that pair and convect at an angle to the
freestream observed for effectively innite wings, nite wings exhibit vortex tubes
that break into two branches near the tip forming double helix structures. One branch
connects with the last vortex; one branch connects with the next vortex. This creates
a long daisy chain of interconnected trailing edge vortices forming a long series of
vortex loops. These symmetric ow elds are shown to persist for nite wings even
to Strouhal numbers more than twice those required to produce asymmetric wakes
on plunging airfoils. Two contributing reasons are discussed for why deected jets
are not observed. First the tip vortex creates three-dimensionality that discourages
vortex coupling. Second, the symmetry of the circulation of the interconnected vortex
loops, which has been conrmed by the experiments, is a natural consequence of the
vortex topology. Therefore, the asymmetry in trailing edge vortex strength previously
observed as characteristic of deected jets cannot be supported for nite wings.
C
2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4891256]
I. INTRODUCTION
Flapping wings are the preferred method of natural propulsion both through the air and through
water. Birds, insects, and sh commonly employ wing oscillations as a means of lift and/or thrust
generation with the kinematic motion altered to evoke different manoeuvres, for example, to loiter or
propel themselves forward. The uid dynamics that govern these ows are of fundamental interest
and have become more important recently due to the growing desire to develop efcient Micro
Air Vehicles (MAVs). Improved understanding of these unsteady ows, and in particular the vortex
shedding mechanisms,
1
is necessary if man-made yers hope to achieve similar performance at
small scales. Natural yers use a large-amplitude low-frequency (low Strouhal number) apping
motion. In this paper, we consider the opposite: small-amplitude high-frequency (high Strouhal
number) motion. The only commonality with nature is the generation of leading-edge and trailing-
edge vortices. The kinematics in our case is relevant to the exploitation of aeroelastic vibrations
for ow control purposes, through excitation by wing oscillations. In fact, our approach is a ow
control method to increase lift, and is only applicable to small aircraft because the frequencies of the
wing structure can provide sufciently high Strouhal numbers.
2
We study rigid airfoils and wings in
small-amplitude forced plunging motion, which mimics bending oscillations. In our previous work
we identied a mechanism of high-lift production due to the bifurcated/asymmetric ows. These
deected wakes can lead to time-averaged lift coefcients higher than any other mechanism.
An important parameter used in the study of plunging wings is the Strouhal number based on
the amplitude, St
A
= 2fa/U

, where f is the plunge frequency, 2a is the peak-to-peak amplitude,


1070-6631/2014/26(7)/071907/15/$30.00 C
2014 AIP Publishing LLC 26, 071907-1
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-2 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 1. (a) Time-averaged velocity showing deected jets for a NACA0012 airfoil at =0

and Re =10 000. The direction


of the deected jet determines the positive or negative lift coefcient and is characterised by asymmetry in the trailing-edge
vortex strength; (b) phase-averaged vorticity (adapted from Cleaver et al.
8
). Reprinted with permission from Cleaver et al.,
J. Fluid Mech. 708, 349376 (2012). Copyright 2012 Cambridge University Press.
and U

is the freestream velocity. It is a dimensionless parameter that describes the ratio between
plunging and freestream velocity. An increase in Strouhal number results in a monotonic rise in
effective angle of attack because it depends on the term: tan
1
(U
pl
/U

) where the maximum plunge


velocity is U
pl
= 2fa. The ability of an oscillating wing to generate thrust is a well-established
phenomenon. A parameter space consisting of apping amplitude, 2a, and the Strouhal number St
A
,
is often considered in order to characterise the wake vortex patterns. However, there is also a strong
inuence of the non-dimensional frequency, St
c
= fc/U

, (where c is the chord length), when there


is separation and a leading-edge vortex. Studies have found that at a sufcient combination of the
amplitude and frequency, a switch occurs in the wake vortex pattern. The staggered array of vortices
that are characteristic of bluff body wakes invert their positions to forma reverse von K arm an Street.
3
Therefore, if the ow is from left to right, the counter clockwise vortices move above the horizontal
line, while the clockwise vortices move below. The wake decit is therefore replaced with a wake
surplus resulting in a jet in the time-averaged ow that is indicative of thrust.
At even higher Strouhal numbers studies have shown that the clockwise and counter-clockwise
trailing edge vortices may begin to shed in pairs and propagate at an angle to the streamwise
axis; thereby describing a form of asymmetric shedding even for zero angle of attack. An example is
shown in Figure 1. This symmetry breaking process was rst observed by Bratt
4
but remained largely
unnoticed until Jones et al.
5
used an unsteady inviscid panel code, and simulated deected jets on a
plunging two-dimensional airfoil. The formation of deected jets was therefore deemed an inviscid
phenomenon and leading edge separation was not considered a prerequisite for its inception. In these
computations, once formed, the direction of the deected jet was constant, and could therefore be
described as stable. It is largely recognised that for these stable deected jets the initial choice of
jet direction, up or down, is determined by the initial conditions,
68
i.e., the starting position and
acceleration time. Godoy-Diana et al.
6
explain that the initial trailing edge vortex entrains uid from
behind, thereby deecting the mean ow which results in a change in the vortex couple trajectory.
Similarly Cleaver et al.
8
showed that the choice of mode is determined by the interaction of the
vortices from the rst two cycles and that this is a very complex process which will depend on
multiple parameters. However, in these studies once established this direction does not change.
Conversely, the water tunnel experiments of Jones et al.
5
revealed that the direction of these
deected vortex couples may somewhat randomly switch between an upward or downward mode
and can therefore be described as unstable in some cases. This instability was not observed in
their inviscid panel method. The viscous solution of Lewin and Haj-Hariri
9
successfully simulated
similar switching for a similar plunging NACA0012 airfoil, implying that the switch may be driven
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-3 Calderon et al. Phys. Fluids 26, 071907 (2014)
by disturbances from the leading edge separation. Experimentally Heathcote and Gursul
10
showed
that switching may occur periodically over a continuous series of cycles with a period on the order of
one hundred plunging period. Likewise Cleaver et al.
11
observed a very similar jet switching period
of around one hundred plunging period for a at-plate airfoil. In this case the geometry was shown
to be crucial, with a NACA 0012 geometry stable deected jets occur whereas with a at-plate
geometry unstable deected jets occur. All of these results would indicate that it is the leading-edge
separation which drives the jet switching phenomenon but deected jets can exist without leading-
edge separation, i.e., their formation is associated with the behaviour of the trailing-edge vortices
(TEVs).
It was shown that dual ow elds (two possible deection angles of the jet) exist even at non-
zero angles of attack up to the stall angle.
8
Observations indicate that deected jets form once the
TEVs reach sufcient strength at high Strouhal numbers. Several studies
5, 6, 8, 1214
reported critical
Strouhal number in the range of St
A
=0.320.45. Critical strength
8
was reported as ||/U

c =2.6
or ||/U
pl
c = 1.85, where is the circulation of the trailing-edge vortices. The rst two criteria,
St
A
and /U

c, cannot be considered universal criteria due to the wide scatter in results both across
different studies and systematic variation within individual studies. The third criterion shows more
promise but needs further validation. Similarly Godoy-Diana et al.
14
consider a condition for TEV
asymmetry on a pitching two-dimensional airfoil, based on the magnitude and direction of the self-
induced velocity of a vortex couple, and its horizontal advection speed relative to the freestream
velocity. The effective phase velocity of the vortex couple denes a parameter in which a positive
value describes a satisfactory condition for deected jets to occur.
In terms of what characterizes the direction of the deected jets Cleaver et al.
8
propose that the
angular velocity of a vortex couple can be used to dene the mode choice. For example, Figure 1
shows a stable upwards deected jet (Mode A left) and downwards deected jet (Mode B right).
For the upwards deected jet the counter-clockwise TEV is the stronger of the two; whereas in the
case of the downward deected jet the clockwise TEV is the stronger of the two. This asymmetry in
the TEV strength describes the direction of the deected jet and was observed for all cases studied
across a range of amplitudes and mean angles of attack. It also explains why dual ow elds are not
observed for larger angles of attack (the angle of attack creates a natural bias in the TEV strength
towards a mode B ow eld).
Naturally these time-averaged jets and the choice of mode have been shown to have signicant
inuence on lift. Cleaver et al.
8
showed that for zero mean angle of attack, an upward deected jet
is associated with signicant positive lift: C
L
3.5, whilst a downward deected jet results in the
opposite, signicant negative lift: C
L
3.5. Similar observations were made by Yu et al.
15
As
deected jets occur at values of St
A
similar to natural yers there is a possibility that the phenomenon
is exploited as a means of lift augmentation in natural ight.
6
However, similar asymmetric wakes
have not been observed for nite wings, which raises the question whether this phenomenon is even
possible in the three-dimensional ows produced by nite wings. Instead of distinct two-dimensional
vortices nite wings exhibit continuous chains of interconnected vortex loops.
1621
As the Strouhal
number is increased beyond about St
A
= 0.35, two oblique jets that are symmetric develop as a
result of two branches of distinct interconnected vortex loops. This symmetric time-averaged wake
is different than the asymmetric wakes for two-dimensional airfoils shown in Figure 1.
In this paper we demonstrate that deected jets do not occur, in the wide range of parameters
studied, for nite wings through direct comparison of an effectively innite wing and nite wing
cases at the same experimental parameters with the same experimental setup. We then present
detailed ow eld measurements for a typical case and suggest two mechanisms that contribute to
the absence of deected jets in nite wing cases.
II. EXPERIMENTAL METHODS
A. Experimental setup
The experiments were performed in a water tunnel at the University of Bath. The Eidetics
R
Model 1520 consists of a free-surface closed-loop system capable of ow speeds of up to 0.5 ms
1
.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-4 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 2. Experimental setup and apparatus for the (a) PIV and (b) volumetric velocimetry system. 1: rotary encoder, 2: motor,
3: force balance, 4: laser, 5: camera, 6: laser sheet/cone, 7: end plate, 8: wing.
The test section is 1520 mm long, 381 mm wide, and 508 mm deep, providing optical access from
beneath, the sides and through a downstream viewing window. A series of honeycomb structures,
located upstream of the test section ensure a turbulence intensity of less than 0.5%.
The experimental rig is placed on top of the test section and drives the vertically mounted,
fully submerged wing in lateral oscillations across the width of the tunnel, see Figure 2. The wing
oscillations are made possible via a crank arm mechanism, powered by an AC 0.37 kW Motovario
three-phase motor with a 5:1 gearbox reduction ratio. The crank armwas designed to be long enough,
relative to the turning radius, to ensure that the wing displacement was sufciently close to that of
a sinusoidal curve. Trailing edge tracking was used to verify that the actual displacement is within
1.25% of the peak-to-peak amplitude of the required sine curve. The motor spin rate and therefore,
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-5 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 3. A schematic of the test section, showing the b/c = 3 rectangular wing. The distance d between the tip of the wing
and the endplate is varied by adjusting the height of the tip endplate, from the base of the test section.
plunging frequency is regulated with an IMO Jaguar Controller and the amplitude is set prior to
the start of the tests and simply dened by the crank arm turning radius. The geometric angle of
attack is also kept constant at = 0

throughout the oscillation to impose a pure plunging motion.


For the majority of the cases studied, Reynolds number is Re = 1 10
4
and the amplitude of the
plunging motion was set to a = 0.15c with the forcing frequency varied in the range St
c
= 02.025,
corresponding to a reduced frequency of k = fc/U

= 06.36. For the lower Reynolds cases,


Re = 0.5 10
4
, the amplitude was still a = 0.15c but the forcing frequency varied in the range
St
c
= 04.050.
The wing has a NACA 0012 cross section, rectangular planform, and a span of b/c =3. In order
to remove free surface effects, an end-plate was placed belowthe free surface, at the root of the wing,
yielding a half-model. For the two-dimensional airfoil cases a further end-plate was placed at the tip
of the wing to create an effectively innite wing (not shown in Figure 2). The maximum blockage
ratio was 3.1%. To investigate the effect of tip spacing (d) and the transition from effectively innite
to nite wing, this end-plate was adjustable so that the height of the tip endplate from the base of
the water tunnel test section could be increased/decreased, see Figure 3.
B. Techniques
1. Force measurements
Force measurements were performed using a two-component binocular strain-gauge balance.
Two sets of four bonded strain gauges were connected to form two Wheatstone bridges with
an excitation voltage of 6.5 V. Signal conditioning was applied using an input instrumentation
AD624 amplier, allowing the signal to be zeroed, amplied and ltered at different stages. The
conditioning card provides two gain stages. The input stage has a maximum gain of 500, and the
output stage has a nely adjustable gain of 5.5410.9. A two-pole low pass Butterworth lter was
set to 30 Hz. Finally, an AD/DA converter prepares the signals for the data acquisition card before
being monitored and stored using an in-house LabView program.
A total of 30 000 data points at a sampling rate of 360 per oscillation cycle were averaged to
generate a mean lift coefcient for a single frequency. This was repeated for various frequencies.
It is important to note that within the instantaneous signal there exists an inertia component. The
average of this inertia force over a complete number of cycles is zero, allowing us to time-average
the instantaneous force to obtain the time-averaged lift force. The uncertainty associated with
these time-averaged lift measurements increases with increasing frequency. For a typical case the
uncertainty of the time-averaged lift coefcient increases from 0.03 at St
c
= 0, to 0.35 at
St
c
= 2.025.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-6 Calderon et al. Phys. Fluids 26, 071907 (2014)
2. Particle image velocimetry
A series of ow eld measurements were taken with a Particle Image Velocimetry (PIV)
system. The TSI Inc. 2D-PIV system comprises of a dual ND:YAG 50 mJ pulsed laser, illuminating
the streamwise plane. The ow was seeded with 812 m hollow glass particles and the illuminated
plane was then captured by a 2MP PowerviewPlus 12bit CCDcamera. The positioning of the camera
and the laser sheet with respect to the test section is illustrated in Figure 2. The end-plate used for
the effectively innite wing, which is not shown in Figure 2, is transparent to allow optical access.
A rotary encoder tted to the worm gear spindle produced 360 pulses per cycle, each within equal
time intervals. Combined with an in-house pulse selection system and a LaserPulse synchroniser
this allowed the laser to be red at any desired phase in the cycle. The pair of laser sheets were
red approximately 750 s apart, depending on freestream velocity and Strouhal number, straddling
two neighbouring camera frames. Phase-averaged measurements were achieved using a total of
140 image pairs taken at the same point in the oscillation. Captured images were then processed
using commercially available software: TSI Insight 3G. The PIV images were processed using a
recursive FFT correlator with a nal interrogation windowsize of 16 by 16 pixels to generate a vector
eld of 199 148 vectors. This gave approximately a 1.2 mm (1.2% of the chord length) spatial
resolution.
The circulation is estimated as explained in Cleaver et al.:
8
the vortex is located using a
vortex identication algorithm with the search centred on the point of maximum absolute vorticity.
The radius of the vortex is then determined by continually expanding from the centre, one spatial
resolution unit at a time, until the increase in the magnitude of circulation is negative or small (<1%).
The circulation calculation itself is done using both line integral and vorticity surface methods. The
agreement between the two methods was generally very good. All circulation results presented
herein are derived from the average of the two.
3. Volumetric velocimetry
Velocity measurements were also performed with a volumetric three-component velocimetry
(TSI V3V) laser system. The ow was seeded with 50 m hollow glass particles, and illuminated
using a dual ND:YAG 200 mJ pulsed laser. Images were captured using three 4MP 12 bit CCD
cameras, xed within a common housing. A pair of laser cones were red 750 s apart, straddling
two neighbouring camera frames. The positioning of the camera and the laser cone with respect to
the test section is illustrated in Figure 2(b). The raw images are converted into usable results through
four processing stages.
Stage 1, image-particle identication: the illuminated particles within each image are located
based on user dened criteria (size, intensity etc.). Stage 2, image-particle pairing: using the location
vectors from the calibration the same particles are located in each of the three images to form
triplets that dene the particles position in 3Dspace. Stage 3, triplet pairing: the triplets are paired
between the two times using the relaxation method rst proposed by Baek and Lee
22
with a tracking
group size of 8 particles, to give randomly located velocity vectors. Stage 4, grid interpolation:
the randomly located vectors are interpolated onto a common grid to generate a series of uniformly
spaced velocity vectors centred within voxels, 8 mmin size, with 50%overlap. The grid interpolation
ultimately produced 25 00035 000 grid vectors within a measurement volume of 152 mm168 mm
136 mm. Phase-averaged measurements are based on 140 vector elds, phase-locked using the
aforementioned rotary encoder. In order to capture a larger region of the ow eld, a series of
measurements were carried out at various spanwise locations and the different volumes merged in
Matlab.
III. RESULTS AND DISCUSSION
A. Force measurements
One of the consequences of oscillating a two dimensional airfoil at sufciently high frequency
is the ability to break symmetry and develop a dual mode ow. Modes correspond to an upward
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-7 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 4. A comparison of the time-averaged lift, at various operating Strouhal numbers, obtained for the effectively innite
wing (labelled 2D) and nite wing (labelled sAR = 3).
or downward deected jet, giving rise to non-zero lift even in the case of a zero mean angle of
attack. Lift coefcient measurements for an effectively innite wing and nite wing are presented in
Figure 4 for = 0

, showing time-averaged lift as function of St


c
. Note that in these experiments
exactly the same wing and setup is used with the only difference being a tip end-plate (located at
a distance d/c = 0.02) added to the effectively innite wing case compared to no end-plate for the
nite wing case. Force data were captured by increasing or decreasing the frequency incrementally,
allowing the ow to settle after each interval. In the increasing case the wing was started from rest
with an initial position of h
i
= 0 and the rst frequency being St
c
= 0.075. In the decreasing case
the wing was started from rest but with the rst frequency being St
c
= 2.025. In the decreasing case
starting position has an inuence
7
and so two are considered: top (h
i
= a) or bottom (h
i
= a) of
the motion.
The effectively innite wing exhibits a bifurcating lift curve. The two branches develop at St
c
= 1.5 and each corresponds to a specic mode. The upper branch corresponds to an upwards
deected jet (see Mode A in Figure 1); the lower branch corresponds to a downwards deected jet
(see Mode Bin Figure 1). Which branch is produced is determined by the initial conditions, for more
detail see Cleaver et al.
8
The relationship between the direction of the lift and deected jet appears
to be contrary to what one would intuitively expect. In the previous paper,
8
we performed a simple
control-volume analysis for the typical measurement volume shown in Figure 1. This revealed that
the momentum ux terms have a small contribution to the lift force. The main contribution comes
from the pressure difference between the upper and lower boundaries, which can be also estimated
by the Bernoulli equation.
Conversely the nite wing displays a single mode of approximately zero time-averaged lift.
Therefore, the dual modes are prevented and symmetric ow is presumably preserved for the nite
wing. As a result the initial condition has no effect to within the bounds of experimental uncertainty.
The only difference between the effectively innite wing and nite wing is the addition of an
end-plate near the wing tip. Therefore, to investigate the transition from single to dual modes the
effect of the distance from the wing tip to this end-plate is presented in Figure 5. Proximity of the
tip end-plate has the effect of inhibiting the tip vortex and thereby forcing a quasi-two-dimensional
ow. Increasing the distance d allows us to gradually transition from the effectively innite wing
(d/c = 0.02) to the nite wing (d/c = 1.58, off the scale). The force measurements presented in
Figure 5 represent time-averaged lift produced by the wing while oscillating at St
c
= 2.025. The
plunging motion is impulsively started fromeither the top (h
i
=a) or bottom(h
i
=a) of the motion
in order to promote the two modes. The transition from dual mode to single mode occurs over a
small range of gap distances (within 0.25c). It is also interesting to note that the variation of the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-8 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 5. The effect of varying the spacing between the wing-tip and end-plate, on the time-averaged lift generated by the b/c
= 3 rectangular wing. The hollow and solid markers represent the wing impulsively started from the bottom and top of the
motion, respectively. The wing is operating at St
c
= 2.025 and Re = 10 000.
lift force is continuous during the transition. As will be discussed later there are two contributing
effects of proximity of the tip end-plate that could explain this transition: rst it inhibits the tip
vortex and second it creates straight vortex tubes instead of vortex rings. Both of these effects could
be signicant in causing the change from single to dual modes as discussed below.
B. PIV measurements
Figures 6(a) and 6(b) show time-averaged velocity magnitude for the effectively innite wing
and nite wing, respectively. The ows are for the same spanwise location (mid-span, z/c = 1.5).
A deected jet is produced by the effectively innite wing, forming either side of the streamwise
FIG. 6. Time-averaged ow showing non-dimensionalised velocity magnitude for the (a) two modes of the effectively
innite wing and (b) the symmetric ow of the nite wing, both taken at z/c =1.5. The wing is oscillated with St
c
=2.025 at
Re = 10 000.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-9 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 7. Phase averaged ow showing non-dimensionalised spanwise vorticity at four phases in the cycle for the mode A ow
of the effectively innite wing (left column) and nite wing (right column), both taken at z/c = 1.5. The wing is oscillated
with St
c
= 2.025 at Re = 10 000.
axis depending on the initial condition. A region of high velocity ow is also seen above or below
the wing, depending on which of the two modes are activated. For the nite wing, symmetric
ow occurs with a single jet existing approximately parallel to the streamwise ow. Moreover, the
wing fails to generate a similar region of high velocity ow above or below the surface. Therefore,
symmetry is preserved both downstream and upstream of the trailing edge. This agrees well with
the time-averaged force measurements which indicate almost zero lift for the nite wing. Also, it is
noted that, because of the relatively large aspect ratio of the nite wing (semi-aspect ratio of 3 or
full aspect ratio of 6), the difference between the effectively innite wing and nite wing is small in
terms of the velocity magnitude in the wake.
Phase-averaged measurements are shown in Figure 7 in the form of spanwise vorticity for the
mode-A of the effectively innite wing and the nite wing. For the effectively innite wing, the
clockwise trailing edge vortex loiters and pairs with the counter-clockwise trailing edge vortex from
the following half cycle. In this instant, the vortex pair advects above the streamwise axis, describing
a mode A vortex pattern. For the nite wing the vortices fail to pair. Instead, the trailing edge
vortices advect along the streamwise axis preserving symmetric ow. Circulation measurements of
the counter-clockwise and clockwise trailing edge vortex for the nite wing indicate a strength of
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-10 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 8. Time averaged ow at the midspan plane of the nite wing, showing non-dimensionalised velocity magnitude. The
wing is operating with St
c
= 4.05 at Re = 5000.
||/(U

c) = 4.15 and 4.16, respectively. By all three strength criteria presented in the introduction
these vortices are sufciently strong (St
A
= 0.61, ||/U

c = 4.16,||/U
pl
c = 2.18) to create
deected jets and dual modes. However, the asymmetry of the vortex strengths observed for deected
jets does not exist. In summary the TEVs are strong enough but balanced and therefore symmetric.
For the effectively innite wing the asymmetry at the trailing-edge is mirrored at the leading-
edge where the upper-surface LEV is stronger than the lower-surface LEV justifying the high
positive lift for this case. For the nite wing, the upper and lower surface leading edge vortices have
a circulation of ||/(U

c) = 2.73 and 2.64, respectively. The similarity in strengths for the nite
wing reects the time-averaged velocity contour as well as the near-zero time-averaged lift.
Measurements were also performed for a higher Strouhal number of St
c
= 4.05, albeit at a
lower Reynolds number of 5000, in order to investigate whether the phenomenon requires a higher
excitation frequency for the nite wing. Reynolds number does not have any effect on the critical
Strouhal number for bifurcation for effectively innite wings
23
over the range of Reynolds numbers:
Re = 250015 000. The velocity contours are presented in Figure 8. This Strouhal number is now
2.7 times greater than that where deected jets begin for the effectively innite wing, and despite
the signicantly stronger trailing edge vortices, there is no signicant deection in the wake and
symmetry is still preserved. It can therefore be concluded that nite wings prevent deected jets.
Further measurements
24
for other planforms and aspect ratios demonstrate the same behaviour.
C. Volumetric velocity measurements
In Figures 9 and 10 we present vorticity iso-surfaces from a top and side view respectively.
The left column is for the effectively innite wing; the right column the nite wing. Each row
represents a different phase, either: top, middle-down, bottom, or middle-up. For the effectively
innite wing the ow eld is clearly asymmetric, compare Figure 9(left) with Figure 7(left). (Note
that measurements were not taken near the wing root). The trailing-edge vortices are remarkably
two-dimensional, forming straight vortex tubes with the only three-dimensional feature being a very
small branch between the two vortices at the tip end-plate. This branch is small in comparison with
the TEVs and is possibly created by the interaction of the TEV pair with the tip end-plate. There
is also some vorticity generation in the small gap between the wing and the end-plate (most visible
in Figures 9(b) and 10(b)). However, despite these interactions with the end-plate, the vortices in
the wake are highly two-dimensional. Conversely the nite wing has a highly three-dimensional
ow eld. Near the root the TEVs form vortex tubes but at a spanwise location these cylinders
break into a double helix structure with two branches. One branch connects with the last vortex;
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-11 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 9. Top view of iso-surfaces dened by non-dimensionalised vorticity magnitude (c/U

= 14, 25, and 50) overlaid


with contours of spanwise vorticity for the effectively innite wing (left column) and nite wing (right column). The wing is
oscillated with St
c
= 2.025 at Re = 10 000.
one branch connects with the next vortex. There is therefore a long daisy chain of interconnected
TEVs forming a long series of vortex loops. To the authors knowledge, this is the rst study that
has revealed the double helix structure in the connection process of vortex loops. The point at which
this double helix structure occurs moves inboard as the TEVs propagate downstream. In addition,
there is a strong tip vortex.
It appears that the three-dimensionality of the trailing-edge vortices inhibit the formation of
vortex couples. For the effectively innite wing, the vortices are highly two-dimensional and form
a couple. This may be a general rule with regard to the formation of vortex couples. Couder and
Basdevant
25
observed the formation of vortex couples in the wakes of cylinders in two-dimensional
ows generated by the soap-lm technique. At the same Reynolds numbers, such vortex cou-
ples have never been observed in three-dimensional wakes. It may be that, in the present case,
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-12 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 10. Side view of iso-surfaces dened by non-dimensionalised vorticity magnitude (c/U

= 14, 25, and 50) overlaid


with contours of spanwise vorticity for the effectively innite wing (left column) and nite wing (right column). The wing is
oscillated with St
c
= 2.025 at Re = 10 000.
two-dimensionality enhanced by the wing oscillations encourages the formation of a vortex couple.
However, the presence of a tip vortex for the nite wing creates three-dimensionality in the vortices.
It is suggested that there are two contributing causes of the prevention of deected jets in nite
wing cases: the tip vortex and vortex loops. The tip vortex is generated around the same time as
the newest trailing-edge vortex, and these are topologically connected. Tip vortices of opposite sign
form at the locations that are mirror images of each other. Hence there is symmetry (with respect to
the free stream direction) in the formation of the tip vortices. We believe that, as the tip vortices are
connected to the trailing-edge vortices, this discourages the formation of vortex coupling, which of
course results in asymmetry with respect to the freestream direction. Another effect of the tip vortex
is the upwash or downwash velocity induced on the spanwise vortices. For example, downwash
created in Phase (b) (in Figures 7 and 9) may produce an adverse strain, which may encourage
separate vortices.
26
The second contributing factor is the vortex loops in the wake. One of the
central observations made by Cleaver et al.
8
for the effectively innite wing is that the direction of
deected jets is characterized by asymmetry in the TEV strength which is dened by the asymmetry
parameter. This asymmetry in TEV strength was observed for all dual mode cases studied. For the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-13 Calderon et al. Phys. Fluids 26, 071907 (2014)
FIG. 11. (a) Simplied isometric view of the vortex chain and (b) top and side view.
vortex loops observed in Figures 9 and 10 this asymmetry cannot be supported. In Figure 11 we
show a simplied schematic of the vortex loop system (for clarity we omit the details such as the
double helix structure and vortex formation process). In addition to the isometric view, we show
two other views, a top view from above the vortex loops and a side view of the section plane shown
with a dashed line in the top view. In this side view, the conventional inverse Karman vortex street
is observed. The loops of opposite sign have the absolute circulations of
1
and
2
, and in the
streamwise plane the total circulations are (
1
+
2
) and (
1
+
2
). Hence, regardless of whether
the loops have the same circulation (
1
=
2
) or differing circulation (
1
=
2
) the total circulation
of the trailing-edge vortices, that are crucial for asymmetric wakes, will always remain equal. Indeed
we have measured the total circulation and found equal magnitudes (||/U

c = 4.15 and 4.16).


Therefore, the symmetry (in terms of magnitude of the circulation analogous to the two-dimensional
case) is a natural consequence of the interconnected vortex loops for the nite wing case.
IV. CONCLUSIONS
Force, PIV, and Volumetric Velocimetry measurements have been presented which show that
nite wings inhibit deected jets. The transition from dual-mode high-lift deected jets to single-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-14 Calderon et al. Phys. Fluids 26, 071907 (2014)
mode near zero-lift ow elds was shown to occur as the distance of the end-plate from the tip is
varied. Atip spacing of just 0.25c was sufcient to enforce a single-mode oweld. In the effectively
innite wing case the trailing-edge vortices form straight vortex tubes that pair and convect at an
angle to the horizontal creating asymmetry as previously reported by other authors. With the tip end-
plate removed to create a nite wing the ow eld becomes highly three-dimensional. Near the root
the TEVs form straight vortex tubes but at a spanwise location these cylinders break into a double
helix structure with two branches. One branch connects with the last vortex; one branch connects
with the next vortex. There is therefore a long daisy chain of interconnected TEVs forming a long
series of vortex loops. In addition, there is also now a strong tip vortex. It is suggested that there
are two contributing reasons for this abrupt change. First the tip vortex creates three-dimensionality
that prevents vortex coupling. It is noted that there is other evidence of formation of vortex couples
in two-dimensional ows and absence of vortex couples when there is some three-dimensionality.
Second, the circulation of the interconnected vortex loops is symmetric due to the vortex topology
of nite wing, and this is conrmed by the circulation measurements. Therefore the asymmetry
previously observed as characteristic of deected jets cannot be supported for nite wings.
ACKNOWLEDGMENTS
The work was sponsored by the Air Force Ofce of Scientic Research, Air Force Material
Command, USAF under Grant No. FA8655-10-1-3093, as well as the Engineering and Physical Sci-
ences Research Council (EPSRC) Studentship, and the RCUK Academic Fellowship in Unmanned
Air Vehicles.
1
I. Gursul, D. J. Cleaver, and Z. Wang, Control of low Reynolds number ows by means of uid-structure interactions,
Prog. Aerosp. Sci. 64, 1755 (2014).
2
J. D. Elldredge and K. Chong, Fluid transport and coherent structures of translating and apping wings, Chaos 20(1),
017509 (2010).
3
M. M. Koochesfahani, Vortical patterns in the wake of an oscillating airfoil, AIAA J. 27, 12001205 (1989).
4
J. B. Bratt, Flow patterns in the wake of an oscillating airfoil (Aeronautical Research Council, London, UK, 1950), Report
No. R & M 2773.
5
K. D. Jones, C. M. Dohring, and M. F. Platzer, Experimental and computational investigation of the Knoller-Betz effect,
AIAA J. 36(7), 12401246 (1998).
6
R. Godoy-Diana, J. L. Aider, and J. E. Wesfreid, Transitions in the wake of a apping foil, Phys. Rev. E 77(1), 016308
(2008).
7
C. L. Liang, K. Ou, S. Premasuthan, A. Jameson, and Z. J. Wang, High-order accurate simulations of unsteady ow past
plunging and pitching airfoils, Comput. Fluids 40(1), 236248 (2011).
8
D. J. Cleaver, Z. Wang, and I. Gursul, Bifurcating ows of plunging airfoils at high Strouhal umbers, J. Fluid Mech. 708,
349376 (2012).
9
G. C. Lewin and H. Haj-Hariri, Modelling thrust generation of a two-dimensional heaving airfoil in a viscous ow, J.
Fluid Mech. 492, 339362 (2003).
10
S. Heathcote and I. Gursul, Jet switching phenomenon for a periodically plunging airfoil, Phys. Fluids 19(2), 027104
(2007).
11
D. J. Cleaver, Z. Wang, and I. Gursul, Investigation of mechanisms of high lift for a at-plate airfoil undergoing small-
amplitude plunging oscillations, AIAA J. 51(4), 968 (2013).
12
J. E. Emblemsvag, R. Suzuki, and G. Candler, Numerical simulation of apping micro air vehicles, AIAA Paper No.
2002-3197, 2002.
13
K. D. von Ellenrieder and S. Pothos, PIV measurements of the asymmetric wake of a two dimensional heaving hydrofoil,
Exp. Fluids 44(5), 733745 (2008).
14
R. Godoy-Diana, C. Marais, J. L. Aider, and J. E. Wesfreid, A model for the symmetry breaking of the reverse Benard-von
Karman vortex street produced by a apping foil, J. Fluid Mech. 622, 2332 (2009).
15
M. L. Yu, H. Hu, and Z. J. Wang, Experimental and numerical investigations of the asymmetric wake vortex structures
around an oscillating airfoil, AIAA Paper No. 2012-0299, 2012.
16
K. D. von Ellenrieder, K. Parker, and J. Soria, Flow structures behind a heaving and pitching nite-span wing, J. Fluid
Mech. 490, 129138 (2003).
17
P. Blondeaux, F. Fornarelli, L. Guglielmini, M. S. Triantafyllou, and R. Verzicco, Numerical experiments on apping foils
mimicking sh-like locomotion, Phys. Fluids 17(11), 113601 (2005).
18
K. Parker, K. D. von Ellenrieder, and J. Soria, Using stereo multigrid DPIV (SMDPIV) measurements to investigate the
vortical skeleton behind a nite-span apping wing, Exp. Fluids 39(2), 281298 (2005).
19
K. Parker, K. D. von Ellenrieder, and J. Soria, Morphology of the forced oscillatory ow past a nite-span wing at low
Reynolds number, J. Fluid Mech. 571, 327357 (2007).
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25
071907-15 Calderon et al. Phys. Fluids 26, 071907 (2014)
20
H. Dong, R. Mittal, and F. M. Najjar, Wake topology and hydrodynamic performance of low-aspect-ratio apping foils,
J. Fluid Mech. 566, 309343 (2006).
21
J. H. J. Buchholz and A. J. Smits, On the evolution of the wake structure produced by a low-aspect-ratio pitching panel,
J. Fluid Mech. 546, 433443 (2006).
22
S. J. Baek and S. J. Lee, A new two-frame particle tracking algorithm using match probability, Exp. Fluids 22(1), 2332
(1996).
23
D. J. Cleaver, Low Reynolds number ow control through small-amplitude high-frequency motion, Ph.D. thesis (Uni-
versity of Bath, 2011).
24
D. E. Calderon, D. J. Cleaver, Z. Wang, and I. Gursul, Wake structure of plunging nite wings, AIAA 2013-2993, 2013.
25
Y. Couder and C. Basdevant, Experimental and numerical study of vortex couples in two-dimensional ows, J. Fluid
Mech. 173, 225 (1986).
26
R. R. Trieling, J. M. A. van Wesenbeeck, and G. J. F. van Heijst, Dipolar vortices in a strain ow, Phys. Fluids 10(1),
144 (1998).
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
150.214.233.165 On: Mon, 01 Sep 2014 11:25:25